Application of cluster analysis to identification flagged power quality measurements in area-related approach

The article presents the use of data mining to power quality issue. The possibility to using cluster analysis as an appreciate tool to realize division into groups representing the measurement period for which aggregated data (within the meaning of PN EN 61000-4-30 standard) contain and do not contain aggregated voltage events is presented. The K-means algorithm sensitivity test to the identification of data containing interruptions, dips, increases and rapid voltage changes was presented. Synchronous measurements carried out in the mining plant network were used for the test data set. The obtained results allow determining the effectiveness of using cluster analysis to identify aggregated data containing voltage events. The area-related in this article is electrical power network of copper mining industry in Lover Silesia.